Potassium deficiency of rice (Oryza sativa L.) has become an increasing problem due in part to inadequate fertilization programs and increasing crop yields. The objectives were to evaluate the influence of K-fertilizer rate on rice grain yield, whole-plant K concentrations at panicle differentiation (PD) and early heading (EH), and selected yield components of rice grown on K-deficient and sufficient soils. Experiments were conducted on three K-deficient soils and one K-sufficient soil. Muriate of potash fertilizer was applied at rates ranging from 0 to150 kg K/ha. Whole-aboveground plant samples were collected at PD and EH to determine K concentrations. At maturity, panicles were collected from plots receiving 0, 75 and 150 kg K/ha and assayed for total spikelet number and percentage of filled spikelets. Mehlich-3 soil-test K ranged from 54 to 74 ppm at K-deficient sites and 150 ppm at the K-sufficient site. As expected, K-fertilizer rate had no influence on grain yield, whole-plant K concentration at EH, spikelet number/panicle, and percent blank spikelets of rice grown in the K-sufficient soil. In K-deficient soils rice receiving no K-fertilizer contained deficient K concentrations at PD (<1.33% K) and EH (<1.00% K) which increased to sufficient levels as K rate increased. Averaged across K-deficient sites, grain yields of rice receiving no K fertilizer were increased by 13-17% from application of 75 to 150 kg K/ha with maximum numerical yields produced by the greatest K rate. Panicle assays, averaged across K-deficient sites, showed that application of 75 and 150 kg K/ha significantly increased total spikelet number from 156 (0 kg K/ha) to 169-175/panicle and percent blank spikelets decreased from 24% (0 kg K/ha) to 20-22%. Data suggest that yield loss from K-deficiency is partially attributed to a reduction in spikelet number/panicle plus an increase in blank spikelets/panicle.